Multi-modal Sensing for Human Motor Understanding: From Clinic to the Sports Field.

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence


Title: "Multi-modal Sensing for Human Motor Understanding: From Clinic to
the Sports Field."

By

Mr. Baichen YANG


Abstract:

Human motor function, from speech and breathing to gait and athletic 
movement, reflects health and physical ability. While assessing motor 
function is important, it is currently confined to clinics and laboratories, 
requiring prescribed tasks and specialized instruments. Moving assessment 
into daily settings promises unobtrusive and low-cost monitoring, but the 
setup differences leave two critical challenges towards complete and reliable 
assessments. Relaxing the task protocol from controlled setting to daily 
conditions causes motor feature entanglement: natural behaviors mix target 
motor features with the task semantics. Replacing laboratory instrumentation 
by mobile sensors causes multi-level domain shift: incomplete physical 
observations make motor-state estimation sensitive to subject and motion 
conditions. Against the entanglement, PDAssess and EasySpiro are developed to 
obtain key disease features out of uncontrolled setups. Against domain shift, 
SnowPose, KneeGuard and ACLGuard are developed to understand kinematics and 
kinetics under highly-dynamic, calibration-free setups. Collectively, these 
systems enable a multi-modal sensing paradigm, bringing clinical motor 
assessment to daily/field setups.


Date:                   Friday, 14 August 2026

Time:                   2:00pm - 4:00pm

Venue:                  Room 3494
                        Lifts 25/26

Chairman:               Prof. Danny Hin Kwok TSANG (EMIA)

Committee Members:      Prof. Qian ZHANG (Supervisor)
                        Prof. Kai CHEN
                        Dr. Wei WANG
                        Dr. Jin QI (IEDA)
                        Prof. Yuanqing ZHENG (PolyU)